Why Are Chip Stocks Falling? The Numbers as of Nov 5, 2026

As of the Nov 5, 2026 close, the PHLX Semiconductor index (SOX), the main US chip index, is 38.4% below its Jun 22 record close of 14,634.72, well past the 20% line usually called a bear market. Reports point to fears of slower AI building, weak company forecasts, financing worries and forced selling, but no single cause is established. This is where each stock stands.

Sorted by fall from the 2026 high, on closes adjusted for splits. YTD starts at the Dec 31, 2025 close and includes dividends, except the index row. * means added by name, explained under How this was measured. Sandisk’s prices are pre‑split: its board approved an 8‑for‑1 split, with split‑adjusted trading expected from Nov 20.

Sandisk, Western Digital, Intel, Astera Labs and Arm closed more than 62% below their highs, and 12 lines more than 50% below. Those closest to their highs are Skyworks, Texas Instruments and TE Connectivity. Only Broadcom, Qualcomm, TE Connectivity and Nvidia are down for the year. Sandisk is 65.5% below its high and still up 239.7%, so “down for the year” and “down from the high” answer different questions. The wider year is on the best‑performing S&P 500 stocks of 2026.

Nvidia fell 9.3% on Nov 5 to $181.60 and is 24.1% below its Oct 6 high of $239.24, and only four lines here sit closer to their highs. Like AMD, it set that high on Oct 6; 19 of the 26 set theirs in June. Its own page is Nvidia stock, and the data behind a decision on it is in should I buy Nvidia stock now.

Yes, on the usual definition of a fall of 20% or more from a high: the chip index is 38.4% below its record close, while the S&P 500 is 5.95% below its own and has not reached even a correction. Bars show each Nov 5 close below its highest 2026 close, and the index highs are also records. The index first closed past −20% on Jul 17, then rose 26.5% from its Jul 29 low to within 9.7% of the record on Oct 6, the day the S&P 500 set its record. It has lost 31.8% since. To get back to the record from here it needs a gain of 62.4%; the arithmetic is in loss recovery math.

Since the data starts in May 1994, the index has had seven earlier falls of 20% or more on closes, each measured from a record until the next record. Five went deeper than this one has so far.

Seven cases do not set a pattern. The index took until Nov 21, 2017 to pass its March 2000 record. Recovery times ran from 55 to 2,266 trading days, and the index’s membership changed over time, so it does not show what happened to the companies it dropped. This is history, not a forecast of this fall’s end or depth.

A few dated events are documented; the reasons linking them to prices are other people’s explanations and are not proven. The sources below mostly address single days around late October and November. We have not sourced a cause for the fall from June to July 29.

Documented: what happened.

• Forecasts. Intel guided on Oct 29 to Q4 revenue of $15.0 to $16.0 billion, after $16.5 billion in Q3, and an adjusted gross margin of 38.0%, down from 44.1%. It fell 28.6% the next day. CNBC reported a disappointing current‑quarter forecast from Qualcomm, which fell 9.5% on Nov 5.
• Votes on data centers. CNBC reported on Nov 4 that voters sided against data centers in 40 of 41 local ballot measures on Nov 3.
• Numbers that did not stop the fall. AMD reported on Nov 3 revenue up 45% to $13.4 billion and data center sales up 97%, and guided Q4 to about $17 billion. It rose 1.0% in the first session after the report, on Nov 4, then fell 17.3% on Nov 5. As of Oct 19, FactSet estimated S&P 500 Q3 earnings growth at 29.0%, or 20.0% without Micron and Nvidia.

Proposed: why it moved prices.

These do not fully agree. One says demand may slow, another says the selling is about positions rather than the business. CNBC’s own summary is that investors were “looking past the numbers” and questioning whether growth can last. Price data show what fell and when, not why. How one link in this chain got here is in why neocloud stocks fell in October and what a neocloud is.

The companies that build and power data centers fell 9% to 15% on Nov 4, 2026, a day the S&P 500 rose 1.58%. The same day CNBC reported that voters sided against data centers in 40 of 41 local ballot measures on Nov 3, and quoted Baird analysts calling the results a “sentiment headwind” for AI cloud companies. We did not check each company’s own news, so the link is reported, not proven. This is where they closed.

Closes are split‑adjusted and price only, from Exent market data. Grouped by what the companies do, not by a measured link to data centers. Chips were not the lead that day: SMH fell 2.75% on Nov 4, then 11.2% on Nov 5, the chip day in the tables above. GE Vernova’s Nov 4 close of $832.83 was below its Nov 2 close of $862.25, undoing its 6.49% gain on Nov 3.

CNBC reported on Nov 4 that of 41 local ballot measures on Nov 3 to restrict or clear the way for AI data centers, voters sided against data centers in 40. That is press attribution. A vote and a fall on the same day is a coincidence in the data until a company or a filed rule links them, and we have not verified any filed proposal.

The next two sessions split the group. From the Nov 4 close to the Nov 6 close, six of the eight fell further, Vertiv most (−11.0%), while NRG (+6.1%) and Vistra (+3.7%) closed above their Nov 4 levels. Over the same two days the S&P 500 fell 4.1% and SMH 8.0%. Nothing here says which way any of them goes next.

Which of these the best‑ranked analysts still back after the drop is locked in Exent Gems.

Chips fell about 2.8 times as much as the S&P 500 that day. The index’s 12.25% fall was the second worst of 8,182 sessions in data back to May 1994; only Mar 16, 2020 (−15.9%) was worse. CNBC reported that Nvidia lost almost $450 billion of market value. All 26 lines fell, from Skyworks (−2.1%) to Sandisk (−22.3%). Even so, the index is still up 27.2% for 2026, against 7.4% for the S&P 500, both price only. Chips’ earlier lead is in did AI and chip stocks beat the S&P 500.

A 3x fund multiplies each day’s move, so it took an 86.4% loss where the unleveraged chip funds lost 33% to 40%. SOXL fell 38.7% on Nov 5 alone. It is also down 2.7% for 2026, while the chip index is up 27.2%.

Two cautions on comparing these. SOXL targets three times the daily move of its own chip index, not the index SMH follows (the MVIS US Listed Semiconductor 25) or the PHLX index above, so SOXX and SMH are comparisons, not its benchmark. And it resets every day, so 3 times a long move means little: 3 times SOXX’s 39.6% would be more than the whole investment. The SEC’s investor bulletin warns that returns over longer periods can differ widely from the multiple. More in how much leveraged ETFs lose in a crash and can you hold TQQQ long term.

No. The gap is the consensus target divided by the close, minus one. A large gap is not a buy signal, and much of it comes from the price falling while targets stand still. We show no gap per stock in the table above for that reason. Among the 16 lines with a target we show, the median gap was +81.6% at the Nov 5 close and +63.8% at the Nov 4 close, on the same targets. Nvidia’s went from +69.5% to +86.8% in one day; AMD’s from +32.1% to +59.7%.

Targets often lag falling prices. Across those 16 lines, 114 of the 238 targets published in the last quarter came in the last month, and for Nvidia and Analog Devices the data shows none in the last month. Older targets were set at higher prices. How much weight targets have earned is in are analyst price targets accurate, and the whole index’s list is in which stocks have the biggest upside to price targets, which uses Nov 4 closes.

This page gives no buy or sell advice, and these figures cannot say whether a fall is over. Data can show how far a stock fell and what was reported. It cannot show the next move. Points investors often check (education, not advice):

What the biggest winners went through on the way is in how to find the next Nvidia. How earnings days have moved the largest stocks is in how Big Tech stocks move after earnings.

The biggest is Nvidia’s report on Nov 17. Nothing here predicts how it, or the stocks, will move; how chip stocks moved after Nvidia’s past reports is in how chip stocks move after Nvidia earnings. We list only dates from company announcements; other chip names on this page have no date we could confirm from the company.

Data. Our calculation on Exent market data: daily closes through Nov 5, 2026, adjusted for splits, including KLA’s 10‑for‑1 on Jun 12 and SOXX’s 3‑for‑1 on Nov 5. “Below 2026 high” compares the Nov 5 close with the highest close since Jan 2, price only, so intraday highs may be higher. We spot‑checked Nov 5 moves against press reports of Nvidia, Qualcomm, AMD, Micron and Applied Materials.

Who is in the list. Every S&P 500 and Nasdaq‑100 member, as listed on Oct 20, 2026, whose company profile says “Semiconductors”: 23 lines. That label also covers chip‑equipment makers, TE Connectivity and Qnity, and we removed none by hand. We added Sandisk, Western Digital and Seagate by name (marked *), which make memory and storage and are labeled computer hardware: 26 lines, 23 in the S&P 500 and 22 in the Nasdaq‑100. Without the three, the median fall from the high is 37.7%. TSMC is in neither index list.

Survivorship. This list holds current members only. Companies that left the indexes are missing, so it can understate how bad a fall was for chip stocks as a whole. It covers a fall still in progress; the highs and order will change.

Targets. Consensus targets from Exent analyst data, pulled Nov 6. We hide a target where it is likely stale or distorted: Sandisk (mean more than 2.5 times the close), Arm (mean more than 15% from the median), and eight lines with fewer than 5 targets published in the last quarter (Astera Labs, Teradyne, Qualcomm, Qnity, Microchip, Monolithic Power, ASML and NXP). Providers calculate consensus differently.

All figures describe a date. This is not a forecast and not a recommendation for any stock. This page is refreshed on the first trading day of each month and the day after Nvidia reports.